The present invention disclosed herein relates to a security monitoring apparatus and method capable of detecting whether a dangerous situation occurs on the basis of a variation pattern of correlation coefficients between sound field spectra according to multi-tone frequencies on the basis of a sound field variation according to a time, and distinguishably detecting intrusion, motion, fire, and ordinary temperature variation situations for each other.
Security sensors used for independently detecting intrusion and a fire situation have been researched and used for a long time. A sensor for detecting intrusion may use one of a passive infrared (PIR) type, an active infrared (IR) type, an ultrasound type, a sound sensing type, a vibration sensing type, and a microwave sensing type, and a sensor for sensing fire may use one of a temperature sensing type, a smoke sensing type, a gas sensing type, and a flame sensing type.
However, as described above, those sensors are used for sensing one of intrusion and fire situations. A security sensor that distinguishes an intrusion situation from a fire situation with one sensor is recently proposed in a type using a principle that a sound field variation pattern measured by using a sound source having multi-tone frequencies is detected, features are extracted according to time and frequency variations, and the extracted features are analyzed.
As an existing technique, Korean patent application laid open No. 2011-0142499 discloses a security system and method through a pattern analysis of sound field variation, which provides a sound field security pattern technique for calculating an average and a deviation of a sound field, detecting a dangerous situation of intrusion on the basis of a variation value (SNR) of an average value of the sound field to an initial deviation of a reference sound field, and issuing an alarm. However, this patent invention has weaknesses in that a time is taken to determine a reference value since measurements are required to be performed twice or more at an initial stage for obtaining the reference sound field deviation, and that reliability of intrusion detection is vulnerable to randomness of a reference deviation obtained by measurements of the limited number of times and inaccuracy of the method of detecting the sound field variation. In addition, such a method has difficulty in distinguishing intrusion, fire, and motion situations for each other.
As another existing technique, Korean patent application laid open no. 2013-0122862 discloses a security monitoring system and method, which provides a technique for distinguishing a fire situation from an intrusion situation by distinguishing feature of the fire situation where a sound field pattern is not varied in shape but moves in a high frequency direction in a temperature increase variation such as fire from the feature of the intrusion situation where the shape itself of a sound field pattern varies. However, the patent invention has difficulty in accurate quantitative determination, since reliability is vulnerable to inaccuracy of the method of detecting a sound field variation with a variation value (SNR) of an average value of the sound field to a deviation of a reference sound field, there are lots of arbitrariness in a method of quantizing a variation degree of a pattern shape and a method of deriving a similarity index through a difference index, and a frequency movement index is variable according to a situation. In addition, there is also difficulty in detection by distinguishing a motion, etc.